Using bumetanide to treat autism appears promising but further clinical trials are needed to confirm this approach

Abstract

In spite of its high incidence, we still have no efficient, approved treatment for autism spectrum disorders (ASD). The latest World Health Organization fact sheet states that ASD affects one in every 160 children worldwide (https://www.who.int/news-room/fact-sheets/detail/autism-spectrum-disorders), which affect one in every 54–66 children in the USA. The development of molecular and genetic tools has helped to identify altered targets in ASD, namely the proteins involved in synaptic transmission, but they have not helped by providing treatment candidates. In our opinion, the massive, and excessive, reliance on the promises to identify novel genetic mutations has not been translated into novel treatments, as we all expected. It is possible that there are more than 800 mutations involved in ASD, which means that we are not short of possible targets, and we have an excess of transgenic mice to investigate. But identifying a mutation in a few children with ASD, and using animal models with that mutation to test treatments, can only be successful if there is an underlying solid concept and experimental evidence for that aim. The failure to develop treatments for Rett or Fragile X syndromes, despite the fact that the mutations were identified decades ago, is vivid illustrations of the difficulty of directly translating these observations to treatments. We have suggested that the main problem is that ASD develops in utero, where it has an impact on brain development that leads to poorly formed and misconnected neurons and networks that remain immature. These immature neurons disturb the generation of oscillations that are essential for behaviour, and this implies that using drugs that selectively block their immature activity is a promising approach. This neuroarchaeology concept1 has been validated experimentally. It illustrates the importance of taking into account the dynamic sequence of events generated by the insult. A recent review pointed out that more than 90% of clinical trials based on preclinical experimental translational research data failed and discussed why that happened.2 The history of using bumetanide to treat ASD is an illustration of how naïve ideas and risky, or nonestablished, concepts can still lead to successful clinical trials. First, experimental data showed that immature neurons have depolarising and excitatory actions due to the inhibitory transmitter Gamma Amino-Butyric Acid (GABA), because of high intracellular chloride levels.3, 4 Then, new experimental data suggested that benzodiazepines had paradoxical effects when GABA had excitatory effects.5 If immature neurons and misplaced neurons in ASD have these properties, then a drug that restores low intracellular chloride levels should attenuate ASD. These are a lot of ifs, but they are worth trying! (Figure 1). We decided to test the effects of bumetanide, a highly selective antagonist of the sodium–potassium–chloride co-transporter (NKCC1), to treat patients with ASD. After a pilot open trial, we carried out a single site and a multiple site double-blind trial6 and both provided positive results. Since then, many independent pilot or double blind randomised trials have confirmed these results7. Bruining et al. published the result of an open-label pilot study and reported that bumetanide efficiently attenuated ASD features in patients with tuberous sclerosis, without having an impact on the occurrence of seizures.8 Other brain imaging and eye tracking trials, carried out by Hadjikhani et al., confirmed the effects of bumetanide, and perhaps, more importantly, provided interesting research directions on the underlying mechanisms and site of actions for the treatment.9 The paper by Fernell et al.10 in this issue of Acta Paediatrica is interesting, as it extends previous observations. Admittedly, it reports an open-label trial of just six children aged 3–14 years old, but there are four important points to highlight here. First, all the children had severe ASD, with an initial Childhood Autism Rating Scale (CARS) total score of above 39. They also had moderate or severe intellectual disability and some degree of language disability. Second, none of the children had any other comorbidities, medical syndromes or epilepsy, suggesting that bumetanide's effects are on ASD symptoms without other interferences. Third, the authors used the 0.5 mg twice-daily dose that we recommended as a result of our previous trials and this is the dose that has been used in most other trials. Fourth, it was positive to see that none of the parents wanted to interrupt the treatment and they all wanted to continue once the trial finished after 3 months. In addition to the classical Clinical Global Impressions (CGI) scale, the authors used the Parental Satisfactory Survey, which compromises a large list of items that enables researchers to determine whether parents consider that there has been an improvement as the result of treatment. This was the first time t